Skip to main content

Table 4 The average localization performance of the DNNs trained by the central HRTF in each cluster

From: Binaural sound localization based on deep neural network and affinity propagation clustering in mismatched HRTF condition

Test set

Central HRTF

 

HRTF 4

HRTF 10

HRTF 13

HRTF 14

HRTF 24

HRTF 33

HRTF 38

Cluster 1

81.22%

51.58%

36.75%

27.10%

43.82%

18.62%

38.86%

Cluster 2

51.22%

84.42%

40.51%

30.18%

39.71%

25.62%

40.09%

Cluster 3

41.64%

46.90%

82.38%

59.60%

66.88%

54.07%

19.11%

Cluster 4

27.49%

31.78%

60.03%

84.38%

41.50%

54.40%

19.70%

Cluster 5

49.83%

40.90%

62.58%

41.49%

83.38%

46.49%

18.33%

Cluster 6

26.67%

26.55%

70.73%

61.75%

57.83%

86.81%

8.03%

Cluster 7

31.97%

26.77%

11.47%

12.51%

15.57%

5.28%

76.83%

  1. Each test set is generated by the HRTFs in the same cluster